Font Size: a A A

Research Of Method About Rapid License Plate Location

Posted on:2011-02-21Degree:MasterType:Thesis
Country:ChinaCandidate:T Q GuoFull Text:PDF
GTID:2178360308462245Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
License plate location technology is the key step to achieve license plate recognition and car identify, which is also one of the important research subject of computer vision and pattern recognition technology applied in intelligent transportation field.Actually, license plate location technology is the specific technology to locate the license plate area in the image of vehicles. However the conditions of collected plates and background images are uncontrollable due to the vehicles images were collected in the natural environment. Random changes and the complex background information have caused great difficulties to the target location. In this paper, the existing method of license plate location made a detailed exposition of these methods for the pixel-based positioning which make the speed of location is slow. Combined with the thought of classification, through the license plate images for training, we can get the classification for detecting license plate.License plate images used the Haar-like features of training license plates, such features are divided into three categories, combinations of these features is characterized by the template. Though the integration image operations transform, the features make that the complexity to access to a regional features values is constant complexity, while it greatly increased speed of operation. In the training process, the sample plates selected carefully and the complex background negative samples, using the Haar-like features of sample, are trained circuit based on Adaboost algorithm, from weak to strong classifier, and then united strong classifiers into a cascade classifier which classified better. In fact, the core idea is that the best elected represent feature and sample of plates identify the license plate as standard.After use of classifiers to locate license plates, it needs to further select the accurate location of license plates and confirm the position information of the license plate, so we adopt pixels horizontal and vertical scanning method for accurate location, and completed the segmentation process of the characters at the same time, which will improve the speed of the system combined two steps.Finally, this paper achieved this method and has been tested, given the accuracy and speed data after the process of the coarse plate location and precise location. By comparing this method to others, it achieved the intended purpose and the effect of rapid license plate location. The method is also a meaningful attempt for Vehicle license plate location.
Keywords/Search Tags:intelligent transportation system, license plate location, Haar-like features, Adaboost algorithm
PDF Full Text Request
Related items